Kelly Wanser is the founder and CEO of Stateless Networks, a company that is leveraging the capabilities of the new generation of network hardware and software to make dramatic improvements to physical networks while helping to virtualize them. Kelly has over two decades of experience — including many patents and awards — in infrastructure, security, data analytics and enterprise software. She is also a strategic advisor to the fusion energy program at Lawrence Livermore National Laboratories National Ignition Facility.
In the past few months, we’ve seen a lot more interest from networking experts in automation solutions — solutions that historically were used by DevOps for server and application management. This may mark a turn towards a simpler approach to software-defined networking (SDN) than more complex fine-grained flow programming (and of course, we’re using the term SDN here more flexibly). When the opportunity to speak with Kelly Wanser, CEO of Stateless Networks, presented itself, I jumped at the chance to understand how she and her team are blazing new trails in the brave new world of automation and next-generation management in networking.
For our readers who haven’t yet heard about Stateless Networks, could you take a few moments to bring us up to speed?
Wanser: Yes! It would be terrific to talk about Stateless with you and SDNCentral. Stateless intends to transform data center networks – and the lives of network teams and application owners – by making network operations systems intelligent. To do that, we’re building a new breed of network operations system that leverages the programmability of the next generation of network hardware and software to automate networks using real-time data and machine learning. We’re replacing polling and CLI with dynamic, intelligent network middleware.
Our flagship product, the Stateless Network Director, is a software system that uses agent interfaces to network elements, real-time analytics, webscale architecture and a unique blend of NetOps and DevOps tooling to automate and manage networks. The Network Director can be used to operate a largely physical network, or a highly virtualized one, and can help networks prepare for, and transition to, virtualization and integrated infrastructure operation.
Our long-term vision is to leverage our webscale approach and big data analytics for integrated infrastructure operation. We think that by starting with the most complex, interconnected part of the infrastructure – the network – we may be well positioned for this.
Stateless Networks is a pre-launch, private company based in San Francisco. Culturally, we combine the creativity and agility of web development with the depth and discipline of hard-core network and systems engineering.
And who are your target customers?
Wanser: We focus on the enterprise data center or “private cloud,” and our initial work has been with financial companies, who are great early adopters of technology where there are market gaps.
Our primary customer is the network organization. We want to make a big difference in the lives of network teams by using automation and intelligence to make the network stronger, faster, more transparent and easier to operate – eliminating 3:00 a.m. calls and weekend fire-drills. We help automate today’s (largely physical) networks safely and provide the visibility and controls needed to adopt virtualization in a risk-managed way.
We also have key stakeholders in cloud and infrastructure management, where network services can be enabled and integrated by Stateless, and champions at the C-level who are prioritizing improvements to network operations as critical to business operations.
Based on recent conversations with some of the largest networking vendors, service providers, and enterprises, we are starting to see a shift away from the flow-based programming flavor of SDN in favor of simpler frameworks. What do you hear in your own conversations with customers?
Wanser: Absolutely. We’re hearing the definition of SDN broadening to include all networking software, especially network operating systems and management systems.
Customers want to invest where software programmability or centralization create a lot of value by accelerating application delivery, improving performance, or reducing operating or capital costs, but without sacrificing existing capabilities. Some SDN solutions are good at one variable, for example — reducing hardware costs — but add cost or risk in other areas. Some are solutions looking for use cases.
It’s exciting now, because customers are actively looking for new solutions to longstanding problems, but bandwidth for experimentation outside of high-ROI/low-risk projects is low, and the SDN flow controller is still a tougher sell. Stateless is able to fill the gap between the desire and vision to have a programmable and automated network, and the hard reality that many SDN solutions built around new controller paradigms aren’t ready for prime-time and scaled deployments.
So, is automation truly the heart of SDN? Are we just talking about DevOps in networking?
Wanser: Is automation the heart of SDN? We think so, yes – automation and intelligence.
SDN can transform a network of static plumbing into an intelligent, unified, system where processes are performed and data is routed in a way that can is dynamically optimized for performance, availability, security, cost or other criteria. In part, what we do – intelligent automation at the management layer – is applying DevOps concepts to networking
Stateless has been called the “gateway drug to SDN” because with our network-intelligence-plus-DevOps approach, we add immediate value to physical network operation, while also adding the visibility and a dynamic operational framework that make it easier to deploy virtualization/centralized control.
It appears that our type of capabilities may be a prerequisite for network virtualization, making Stateless earlier in the order of operations.
”Gateway drug to SDN” — that’s definitely an analogy I won’t forget anytime soon. And that seems pretty apt too. Can you expand on how networking automation is different? Specifically, how is Stateless different from the Chefs, Puppets, Salts, Ansibles of the world? What’s special about Stateless?
Wanser: DevOps tools like Chef, Puppet, Salt and Ansible were developed for server automation. While the server is not trivial, servers can be managed independently — pulled in and out of rotation — without impacting other elements and with just a few key pieces of feedback data (CPU utilization, etc.) from a node.
The network, however, is an interconnected system in which a change can impact an array of connected elements and upstream applications. Automating change on a large, complex, interconnected system requires real-time data from all across the system (network and attached hosts and applications), with extensive logic regarding the conditions the network needs to maintain and the dependencies required to maintain them.
Server DevOps tools can push changes to network elements, but, lacking system-wide intelligence, have network operators flying blind. That is the gap. And, because of the interconnected nature of the network, and its role of “keeping the lights on” for all applications, network operators face greater risks, with a much lower margin for error. Developers using DevOps tools can experiment in their own sandboxes. Network operators are operating on a live system in which changes, or errors, can have large, inadvertent and incredibly costly consequences. For DevOps to move into networking, the tools need to support network operations’ requirements for comprehensive feedback and risk mitigation, rather than asking that they become more risk-tolerant.
But, it’s a large, specialized area. The network is data-intensive, subject-matter-rich, speed-sensitive and must be modeled as a system, so the kind of data architecture and analytics required are quite a leap from server tools.
What kind of use cases are the most dominant for Stateless, and what kind of benefits and ROI are your early customers seeing?
Wanser: The first generation of the Stateless Network Director focuses on provisioning and change management automation use cases. Our sweet spot is real-time dynamic conditional logic (including operational status and policy enforcement) applied to change.
We see very high ROI in labor reduction — 70 to 80 percent less labor spent in hardware provisioning, maintenance and incident response.
But the greater gains are likely to come from the acceleration of application deployment, from weeks or days to hours, and the reduction of outages by 70 percent or more. We have more to quantify as we gain experience. And, our projects today are for physical networks, but with use cases for visibility into hypervisors and network overlays.
Who are Stateless’ natural partners?
Wanser: Stateless is a flexible intelligence and automation layer that sits between the physical and virtual network, and between network and cloud operations systems. Our natural partners are network hardware and software vendors like Arista, Cisco, Cumulus, Juniper, HP, Dell and Brocade; virtualization, orchestration and cloud management vendors like VMWare, Opscode and various Openstack vendors; and infrastructure operations vendors like HP, BMC and IBM.
The big infrastructure systems vendors are interesting, because they have huge, lucrative businesses in operations systems for networks- but these systems were built for the polling/CLI model of operation, and are not architected for the programmability, the scale, or the virtual and physical nature of the new network platforms, and are likely to be obsoleted by the new capabilities.
Stateless is also a natural partner for integrators and VARs who are deploying next-generation networks with their customers and providing services around automation and new network operations processes.
What’s been the biggest surprise for you in the last 12 months?
Wanser: There have been a few! The first was the speed of traction we gained with the first companies we talked to – major financial institutions – and then, the level of interest from almost everyone else. We found ourselves in a gap in which customers are trying adapt and integrate a combination of legacy network tools and newer server tools with homegrown data stores and logic, in order to automate networks in a way that leverages the programmability of the new fabric systems. We were surprised at our timing.
With that, we were also surprised by the C-level attention and support we received. We had not expected network operations to be visible as high in the enterprise, or for Stateless to have active support from prominent, senior executives in major institutions.
Apart from that, we built a cool consumer-grade UI to accompany our system, but expected primary interest to be headless operation via our APIs. We were pretty surprised (all of us but the front-end team) at the level of interest in our UI.
How do you see automation and SDN evolving over the next 12 months?
Wanser: Automation is coming to the network. It started with web and cloud operators building custom systems and other enterprises automating around the edges by combining legacy network and newer server tools with some cut-and-paste human integration against specific use cases.
In the next 12 months, most vendors will support open, programmable hardware operating systems, and their adoption will increasingly include systems like Stateless for automating the physical network and dynamic awareness of virtual elements.
As these systems leverage data to drive actions dynamically, the operation of the the network will look more like operation of systems that serve ads in Facebook, recommend TV shows in Netflix, or order materials and parts for Apple’s supply chain, intelligently serving up the best content — in this case network services optimized for policy, security, performance, and cost — on demand. With this in place, we think network virtualization — the sort of centralized traffic flow functions originally associated with SDN — will accelerate very rapidly.
And finally, how would I know whether I can benefit from Stateless’ solutions and what’s the best way to get started?
Wanser: If you operate an enterprise data center and plan to increase automation or implement virtualization, you should contact Stateless. If you are in a refresh cycle, expanding or building greenfield, you should contact us to accelerate your deployment, and leverage new capabilities. We make it easier to work with newer vendors’ capabilities out of the box — with Arista, for example, our UI and tools make it easy to take advantage of their advanced capabilities without custom development. Start by contacting us at email@example.com or visit our website at www.statelessnetworks.com.
If you are a top network, software or data engineer, and have the wit, creativity and courage to apply your skills in a new way, you should also contact us.
Thank you, Kelly!
Check out more Interviews on SDNCentral:
- A Chat with Thomas Nadeau and Ken Gray, Authors of O’Reilly’s SDN Book
- Stateless Networks, the ‘Gateway Drug’ to SDN? A Chat with Kelly Wanser, CEO
- SDDC13 Speaker Viewpoint: Software Defined Data Center — A Regulated Enterprise Perspective, by Andy Brown, Group CTO, UBS
- Software-Defined Data Center Symposium Q&A: Greg Ferro of Packet Pushers and Ethereal Mind
- SDDC Q&A: Ivan Pepelnjak and Brent Salisbury
- Interview: OpenDaylight’s David Meyer — After ‘Breaking Dawn’
- NEC, Nuage and Nutanix on the Inevitability of the Software-Defined Data Center (#SDDC13)
- Featured Interview: 6WIND Targets NFV’s Performance Bottlenecks
- Margaret Chiosi on Why AT&T Wants NFV
- SDDC Interview: Tech Field Day’s Stephen Foskett and Tom Hollingsworth